eb-html.wok

#!/usr/bin/wok -f

/** eb-html.wok
  * 
  * Usage:
  *   java -jar wok-0.1.0.jar -f eb-html.wok -v src=0 -v dst=1 -v update=true -v@rawstr dic=path_to_epwing_dir -v@rawstr ebmap=path_to_ebmap_file input > output
  *
  * Option:
  *   -v src:           The index number of the source column.
  *   -v dst:           The index number of the destination column.
  *   -v@rawstr dic:    Path to a directory containing Epwing's CATALOGS file.
  *   -v@rawstr ebmap:  Path to a EBWin3/4 map file corresponding to the EPWING dictionary specified by `dic`.
  *   -v update:        Whether or not update all cells of the column specified by `dst`.
  */

FS = '\t'
FQ = Quote.Min
OFS = '\t'
OFQ = Quote.All

val eb = "ebquery-0.3.1.jar"

if (eb nonExistent) 
  eb write Resource.fromURL("https://bitbucket.org/rubyu/ebquery/downloads/ebquery-0.3.1.jar").bytes

def query(s: String) = Seq("java", "-Dfile.encoding=UTF-8", "-jar", eb, "-d", dic, "-f", "html", "-m", "en,tx,sb,sp,em,dc,kw,ls,ec", "--ebmap", ebmap, s).!>.string

In { _ 
  .filter (_ isDefinedAt src)
  .map (_.padTo(dst+1, ""))
  .map (row => if (update || row(dst).isEmpty) row.updated(dst, query(row(src))) else row)
  .foreach (row => println(row: _*))
}